> On Feb. 15, 2017, 3:08 p.m., Robert Nettleton wrote:
> > ambari-server/src/main/resources/stacks/HDP/2.2/services/stack_advisor.py, 
> > line 1032
> > <https://reviews.apache.org/r/56706/diff/1/?file=1634453#file1634453line1032>
> >
> >     Should the number of shards be hard-coded here?

Unfortunately at this point we have no way of knowing the number of shards, as 
the user who installed the portal may have altered the name of the service / 
audit logs collecitons. Setting it to 2 ensures that it will work as there are 
at least 2 shars. The user who installs the logfeeders have the option to set 
up the proper value, just like they have the option to set up the proper 
collection names and split intervals.

In the future we plan that the portal should create two znode ssomething like 
this:

/infra_solr/service_logs_info.json:
{
  "collection_name": "hadoop_logs",
  "number_of_shards": "3",
  "split_interval: "!5"
}

/infra_solr/audit_logs_info.json:
{
  "collection_name": "audit_logs",
  "number_of_shards": "2",
  "split_interval: "!5"
}

Then the Log Feeders will get all these infos from ZooKeeper irrelevant of the 
portal location (same cluster, different cluster), and then the logfeeder-only 
install could hide all the logsearch properties.


- Miklos


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/56706/#review165708
-----------------------------------------------------------


On Feb. 15, 2017, 1:09 p.m., Miklos Gergely wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/56706/
> -----------------------------------------------------------
> 
> (Updated Feb. 15, 2017, 1:09 p.m.)
> 
> 
> Review request for Ambari, Oliver Szabo and Robert Nettleton.
> 
> 
> Bugs: AMBARI-20022
>     https://issues.apache.org/jira/browse/AMBARI-20022
> 
> 
> Repository: ambari
> 
> 
> Description
> -------
> 
> Allow the choice to install Log Search without a portal, i.e. to install only 
> Log Feeders which load the entries into an external solr which are viewable 
> via an external portal.
> 
> 
> Diffs
> -----
> 
>   
> ambari-server/src/main/resources/common-services/LOGSEARCH/0.5.0/metainfo.xml 
> 245a94d 
>   
> ambari-server/src/main/resources/common-services/LOGSEARCH/0.5.0/package/scripts/params.py
>  fecd802 
>   ambari-server/src/main/resources/stacks/HDP/2.2/services/stack_advisor.py 
> 5e0fa5e 
> 
> Diff: https://reviews.apache.org/r/56706/diff/
> 
> 
> Testing
> -------
> 
> Tested two local Vagrant clusters.
> 
> 
> Thanks,
> 
> Miklos Gergely
> 
>

Reply via email to